Research · 2 min read
After Dark, Weather and the Moon Change Little
The midnight sensitivity holds under environmental controls; the moon shows no effect and weather very little.
The after-dark question got a famous answer in this library: midnight timestamps can flip the sign. This article adds the environment around it — season, weather, and the moon — and finds the moon does nothing, weather does very little, and the darkness effect stays sensitive to how you treat time.
Our analysis of 116,846 records over 1,096 days, july 1, 2023–june 30, 2026.
Does the after-dark pattern survive weather, season, and moon controls?
Across 116,846 Seattle property-offense records, the night/day ratio is 1.080 as recorded but falls to 0.899 once 10,168 exact-midnight records are excluded. Full-moon days are within 3% of other days, and weather coefficients are near zero—so darkness remains a timestamp-precision question, not a scheduling rule.
The numbers behind the answer
Selected measures only. Denominators and interpretation stay attached so the headline cannot stand alone.
1.080
Night/day ratio as recorded
Night records were 8% more frequent than daylight records when exact-midnight timestamps were kept.
0.899
Ratio excluding midnight
Drop 10,168 exact-midnight records and night falls 10% below daylight.
1.03
Full-moon day vs other days
Daily record counts on full-moon days were within 3% of other days — a null result.
The midnight effect survives weather and season controls
Across 116,846 property-offense records over three years, the raw night/day comparison is 1.080 — night looks busier. Remove the 10,168 records stamped at exactly 00:00 and the ratio drops to 0.899: night is 10% quieter. The reversal that defined the earlier article holds in a longer, weather-joined window.
That does not mean darkness reduces crime. It means a dataset where 8.7% of timestamps are exact midnight cannot decide the question on its own.
Evidence visual
Night/day rate ratio by timestamp treatment
Dropping exact-midnight timestamps reverses the comparison.
Weather barely moves it; the moon does not at all
We fit a negative-binomial model of daily counts with seasonal harmonics, weekends, moon illumination, and local weather. The moon coefficient is effectively zero, and the full-moon comparison lands at 1.03 — consistent with a large literature that keeps finding no lunar effect on crime.
Temperature and precipitation enter with tiny, opposite coefficients. Weather is a confounder worth controlling, not a headline driver. The seasonal terms carry more signal than either.
Schedule from local hours, not from dark or moon
Nothing here supports scheduling coverage around sunset or moon phase. The defensible input remains the property's own hourly pattern — closing times, deliveries, occupancy — reviewed against current data rather than a slogan.
- Separate exact times from windowed times.
- Compare clock hour and time relative to sunset.
- Control for season and weekday before claiming an after-dark effect.
- Do not schedule from moon phase.
- Recheck the schedule periodically.
Questions property teams ask
Does this prove darkness causes offenses?
No. It is descriptive and the night/day comparison reverses with timestamp handling.
Does the full moon affect property crime?
In this data, no. Full-moon days were within 3% of other days.
Does weather explain the pattern?
Very little. Temperature and precipitation coefficients are near zero; seasonal structure carries more signal.
Should coverage start at sunset?
This analysis does not support that rule. Use the property's own hourly and operating patterns.
Our methods, limits, and sources
How we calculated this
This is original 911 Sentinel research — we gathered the records, ran every calculation below, and published the aggregate dataset.
We joined 116,846 Seattle property-offense records (2023–2026) to local solar boundaries, NCEI hourly weather, and computed moon phase, then fit a negative-binomial daily model.
- We classified each record as night or day using a sunrise equation validated against USNO.
- We built a daily panel with seasonal harmonics, weekends, moon illumination, and weather.
- We fit a negative-binomial model and computed a full-moon comparison.
- We repeated the night/day comparison with exact-midnight records retained and excluded.
What this analysis cannot establish
- Association does not establish that darkness causes offenses.
- Recorded start time may be estimated or windowed; exact-midnight values are imprecise.
- One central solar coordinate is used for Seattle.
- Weather coverage depends on NCEI availability and is not a per-incident measurement.
- The full-moon test is designed for a possible null and reports one.
Sources
The raw records come from the sources below; the study design, analysis, charts, and conclusions are our own.
Related questions and practical guides
New coverage — intake open
Test the hours that matter at your property
A property review can combine public time patterns with lighting, deliveries, access activity, vacancy, and current operating hours.